Method and apparatus for controlling one or several variables depending on several control inputs

Abstract

In order to control a variable of a control system depending on a plurality of control inputs to minimize an error function, one control input at a time is, at regular intervals, varied step by step in the same direction until the sign of the differential quotient with respect to time of the absolute value of the error between a predetermined desired value and the actual value of the variable has changed from minus to plus, whereupon the same control input is varied in steps in the opposite direction, with the remaining control inputs varied in the same manner cyclically and sequentially after a respective predetermined number of such sign reversals.

What is claimed is: 
     
       1. Apparatus for controlling at least one variable in dependence on a plurality of control inputs to minimize an error function comprising: (a) a plurality of bi-directional digital counters, the output signals of which form the control inputs;   (b) a clock generator;   (c) a circulating shift register;   (d) means to sequentially and cyclically couple the output of said clock generator to the count inputs of said bi-directional digital counters in dependence on the output signals of said circulating shift register;   (e) means for forming the absolute value of the difference between a predetermined target value and the actual value of the control quantity;   (f) a differentiator having the output of said means for forming as an input the output;   (g) a zero crossover detector having the output of said differentiator as an input;   (h) a counter which operates repetitively from zero to a definite value having the output of said detector as an input;   (i) means to reset said counter and to advance said shift register upon reaching said definite value; and   (j) a flipflop having the output of said detector as a trigger input, the output of said flipflop coupled to the inputs of said bi-directional digital counters determining the direction of counting.   
     
     
       2. A method for controlling at least one variable of a controlled system in dependence on a plurality of control inputs to the controlled system to minimize an error function, comprising automatically carrying out the steps of: (a) developing an error signal, in summing means, which is the difference between the actual value of said at least one variable and a desired value of said at least one variable;   (b) finding the absolute value of said error in absolute value forming means;   (c) differentiating the absolute value so found to obtain a differential with respect to time in differentiating means;   (d) always varying one of the control inputs to the control system in steps at regular time intervals in the same direction until the sign of said differential with respect to time has changed from minus to plus using minimum value control means;   (e) thereupon varying the same control input in steps in the opposite direction using said minimum value control means;   (f) repeating said steps of varying in one direction and the other using said minimum value control means; and   (g) varying the remaining control inputs using said minimum value control means in the same manner cyclicly and sequentially after a respective predetermined number of said sign reversal, using said minimum value control means.   
     
     
       3. The method according to claim 2 and further including the step of finding the difference between said error signal and a pre-determined limit value in additional summing means prior to finding said absolute value. 
     
     
       4. In a raw material mill, a method of controlling at least one variable in dependence on a plurality of control inputs to the mill to minimize an error function comprising the steps of: (a) automatically simulating actual values of said variable using a fast mill model simulator having the control inputs to the mill as its inputs;   (b) determining in first summing means a first set of error signals which are the difference between desired values of said variable and actual values of said variables;   (c) deriving, from said first set of error signals, target values of said variables and superimposed modulus controls;   (d) comparing said simulated actual values with said target values in second summing means to obtain a second set of error signals;   (e) determining the absolute value of said second set of error signals in absolute value forming means;   (f) finding the differential with respect to time of said second set of different signals in differentiating means;   (g) always varying one of the control inputs to the control system in steps at regular time intervals in the same direction until the sign of said differential with respect to time has changed from minus to plus using minimum value control means;   (h) thereupon varying the same control input in steps in the opposite direction using said minimum value control means;   (i) repeating said steps of varying in one direction and the other using said minimum value control means; and   (j) varying the remaining control inputs using said minimum value control means in the same manner cyclicly and sequentially after a respective predetermined number of said sign reversals using said minimum value control means.